BPF List
 help / color / mirror / Atom feed
From: Eduard Zingerman <eddyz87@gmail.com>
To: Andrii Nakryiko <andrii@kernel.org>,
	bpf@vger.kernel.org, ast@kernel.org,  daniel@iogearbox.net,
	martin.lau@kernel.org
Cc: kernel-team@meta.com
Subject: Re: [PATCH v2 bpf-next 2/4] bpf: handle bpf_user_pt_regs_t typedef explicitly for PTR_TO_CTX global arg
Date: Tue, 13 Feb 2024 18:40:06 +0200	[thread overview]
Message-ID: <e3b68a899b8ade18addd198d6f33dcbbed473c3c.camel@gmail.com> (raw)
In-Reply-To: <20240212233221.2575350-3-andrii@kernel.org>

On Mon, 2024-02-12 at 15:32 -0800, Andrii Nakryiko wrote:
> Expected canonical argument type for global function arguments
> representing PTR_TO_CTX is `bpf_user_pt_regs_t *ctx`. This currently
> works on s390x by accident because kernel resolves such typedef to
> underlying struct (which is anonymous on s390x), and erroneously
> accepting it as expected context type. We are fixing this problem next,
> which would break s390x arch, so we need to handle `bpf_user_pt_regs_t`
> case explicitly for KPROBE programs.
> 
> Fixes: 91cc1a99740e ("bpf: Annotate context types")
> Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
> ---

Nit: same could be achieved w/o special casing kprobes by looking
     if typedef's type is named before skipping, e.g. as below.
     But I do not insist, probably good as it is as well.

---

diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index f0ce384aa73e..830635b37fa1 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -907,11 +907,9 @@ bool btf_member_is_reg_int(const struct btf *btf, const struct btf_type *s,
 }
 
 /* Similar to btf_type_skip_modifiers() but does not skip typedefs. */
-static const struct btf_type *btf_type_skip_qualifiers(const struct btf *btf,
-						       u32 id)
+static const struct btf_type *__btf_type_skip_qualifiers(const struct btf *btf,
+							 const struct btf_type *t)
 {
-	const struct btf_type *t = btf_type_by_id(btf, id);
-
 	while (btf_type_is_modifier(t) &&
 	       BTF_INFO_KIND(t->info) != BTF_KIND_TYPEDEF) {
 		t = btf_type_by_id(btf, t->type);
@@ -920,6 +918,12 @@ static const struct btf_type *btf_type_skip_qualifiers(const struct btf *btf,
 	return t;
 }
 
+static const struct btf_type *btf_type_skip_qualifiers(const struct btf *btf,
+						       u32 id)
+{
+	return __btf_type_skip_qualifiers(btf, btf_type_by_id(btf, id));
+}
+
 #define BTF_SHOW_MAX_ITER	10
 
 #define BTF_KIND_BIT(kind)	(1ULL << kind)
@@ -5695,9 +5699,25 @@ bool btf_is_prog_ctx_type(struct bpf_verifier_log *log, const struct btf *btf,
 	const char *tname, *ctx_tname;
 
 	t = btf_type_by_id(btf, t->type);
-	while (btf_type_is_modifier(t))
-		t = btf_type_by_id(btf, t->type);
-	if (!btf_type_is_struct(t)) {
+
+	/* Skip modifiers, but stop if skipping of typedef would
+	 * lead an anonymous type, e.g. like for s390:
+	 *
+	 *   typedef struct { ... } user_pt_regs;
+	 *   typedef user_pt_regs bpf_user_pt_regs_t;
+	 */
+	t = __btf_type_skip_qualifiers(btf, t);
+	while (btf_type_is_typedef(t)) {
+		const struct btf_type *t1;
+
+		t1 = btf_type_by_id(btf, t->type);
+		t1 = __btf_type_skip_qualifiers(btf, t1);
+		tname = btf_name_by_offset(btf, t1->name_off);
+		if (!tname || tname[0] == '\0')
+			break;
+		t = t1;
+	}
+	if (!btf_type_is_struct(t) && !btf_type_is_typedef(t)) {
 		/* Only pointer to struct is supported for now.
 		 * That means that BPF_PROG_TYPE_TRACEPOINT with BTF
 		 * is not supported yet.

  reply	other threads:[~2024-02-13 16:40 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-12 23:32 [PATCH v2 bpf-next 0/4] Fix global subprog PTR_TO_CTX arg handling Andrii Nakryiko
2024-02-12 23:32 ` [PATCH v2 bpf-next 1/4] bpf: simplify btf_get_prog_ctx_type() into btf_is_prog_ctx_type() Andrii Nakryiko
2024-02-12 23:32 ` [PATCH v2 bpf-next 2/4] bpf: handle bpf_user_pt_regs_t typedef explicitly for PTR_TO_CTX global arg Andrii Nakryiko
2024-02-13 16:40   ` Eduard Zingerman [this message]
2024-02-13 17:02     ` Andrii Nakryiko
2024-02-13 17:08       ` Eduard Zingerman
2024-02-13 18:12         ` Andrii Nakryiko
2024-02-13 18:48           ` Eduard Zingerman
2024-02-13 18:59             ` Andrii Nakryiko
2024-02-12 23:32 ` [PATCH v2 bpf-next 3/4] bpf: don't infer PTR_TO_CTX for programs with unnamed context type Andrii Nakryiko
2024-02-12 23:32 ` [PATCH v2 bpf-next 4/4] selftests/bpf: add anonymous user struct as global subprog arg test Andrii Nakryiko
2024-02-13 12:51 ` [PATCH v2 bpf-next 0/4] Fix global subprog PTR_TO_CTX arg handling Jiri Olsa
2024-02-13 16:39 ` Eduard Zingerman
2024-02-14  2:50 ` patchwork-bot+netdevbpf

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=e3b68a899b8ade18addd198d6f33dcbbed473c3c.camel@gmail.com \
    --to=eddyz87@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=kernel-team@meta.com \
    --cc=martin.lau@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox